The Ethical Perils of Predictive Policing
Abstract
This paper addresses the history and current state of predictive policing in the United States and examines law enforcements application of big data analytics and artificial intelligence/machine learning through a series of ethical lenses. At its core, predictive policing distills humans and human behavior into data points and those data points are weighted by one set of people with real-life impact on the livelihood and freedoms of other people. This reduction to data gives the users, the police organizations, and the city and state governments the sensation of high-fidelity, neutral, immutable, and actionable data. But assuming programmatic outputs are free of prejudice is dangerous and ends up codifying and entrenching bias. Organizations ranging from domestic police departments to the Department of Defense (DoD) seek to increasingly leverage big data. As they do so, they must change their paradigm from presuming data is neutral and accurate to expecting and addressing bias and gaps. In this way, organizations like police departments, predictive policing program companies, and the DoD more effectively work towards a better and more ethical application of technology and data and prevent systemic bias from being encoded as the new normative standard.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 25, 2022
- Accession Number
- AD1181349
Entities
People
- Deborah Gaddis
Organizations
- Naval War College